AI signature verification, recorded on the blockchain
Verify every
signature.
Prove every result.
Kumpirma checks a handwritten signature against the signatory's own reference signatures, explains the result in plain language, and records it on the blockchain with a certificate anyone can verify.
- 20 reference signatures per signatory
- 5 independent evidence checks
- QR-verifiable certificates
Authorized signatory
- Signature detected
- Print and lines removed
- Compared with 20 reference signatures
Reference Signatory
Juan dela Cruz
Compared with 20 reference signatures. 80% and above is genuine.
Overall signature shape
Matches: closer to the references than nearly all forgeries in testing.
Proportions and rhythm
Matches: closer to the references than nearly all forgeries in testing.
Stroke direction and slant
Inconclusive: both genuine signatures and forgeries fell in this range in testing.
How it works
From a scanned document to a verified record
Five guided steps. You stay in control of what is checked, and every result comes with the evidence behind it.
Step 01
Upload the scanned document
Upload the scanned copy of the signed document and enter the date it was signed.
PNG, JPEG or PDFUp to 25 MBStep 02
Review the detected signatures
Every signature on the page is found for you. Confirm each one, add any that were missed, rotate a tilted one, and choose whose signature it should be.
Add ManualRotateAssign signatoryStep 03
Check the cleaned signature
Printed text, lines and stamps are removed so only the pen strokes remain. Keep the cleaned version, use the original, or erase leftovers yourself.
Cleaned or originalManual EraseStep 04
Compare with the signatory's references
The signature is measured against the signatory's own enrolled signatures on five kinds of evidence, allowing for how their signature naturally varies.
20+ references5 evidence checksStep 05
Receive a result you can prove
You get a verification score with the reasons behind it. The result is recorded on the blockchain, and a PDF certificate with a QR code is ready to download.
80% and above is genuineQR certificate
Technology
The technology behind every result
Models built for real, imperfect documents, and a blockchain record that makes every result tamper-evident.
Finds every signature on the page
Signatures are located automatically, even on crowded letters and forms, so nothing has to be cropped by hand.
Lifts the signature off the paper
Printed text, ruled lines, stamps and scanner noise are separated from the ink, so only the signer's own strokes are compared.
Judged against the signer's own variation
No two genuine signatures are identical. Kumpirma learns how each signatory's signature naturally varies, then checks shape, proportions, stroke direction, stroke detail and construction against that range.
A record no one can quietly change
Each result is fingerprinted and anchored to a smart contract, with its record kept on IPFS. Any later change to a result is detectable.
Who it's for
Built for both sides of a signature
Verifiers get answers they can act on. Signatories get a say in how their signature is protected.
Check a signature before you rely on it
For anyone who has to confirm that a signature is genuine before accepting a document.
- Upload a scanned document and review each detected signature
- Get a verification score with plain-language confidence factors
- Download a PDF certificate that anyone can check by QR code
- Keep every result and its evidence in your history
Protect the signature that is yours
For people whose signatures appear on documents that others need to trust.
- Enroll once with 20 natural signatures on a printable form
- Your own natural variation becomes the standard
- See every verification made against your signature
- References are locked for 90 days after each update
Role-based access
Verifiers, signatories and administrators each see only what their role allows.
Private by design
Personal data is handled in line with the Data Privacy Act of 2012 (RA 10173).
Tamper-evident history
Every result is anchored to the blockchain, so a later change is detectable.
The team
The people behind Kumpirma
A four-person team covering the AI models, the blockchain record, the web experience and quality assurance.
Kurt Joshua Cayaga
AI Developer
Lead Programmer
Emiel James Escuzar
Blockchain / Cloud Engineer
Charles Platon
Web Developer / UI/UX
Raziel Lloyd Sevilla
Quality Assurance
Lead ResearcherFAQ
Frequently asked questions
What verifiers and signatories most often ask about using Kumpirma.
Know before you trust a signature.
Create an account as a verifier to check documents, or as a signatory to enroll your own signature.